Anal. Chem. 1994,66, 719-724
Capillary Gas Chromatographic Introduction of Environmental Compounds into a Trochoidal Electron Monochromator/Mass Spectrometer James A. Larambe and Max L. Delnzer' Department of Agricultural Chemistry and Environmental Health Sciences Center, Oregon State University, Corvallis, Oregon 9733 1
A gas chromatograph was interfaced to an electron monochromator/quadrupole mass spectrometer. The new system was tested for the analysis of environmental compounds. Detection sensitivity for hexachlorobenzene (HCB) through the gas chromatograph was 5 pg or better, and a mass-resolved molecular ion cluster for this compound on the fly was achieved with 10.8 ng of sample. An ion chromatogram was obtained using 45 ng of Aroclor 1254, and the extract from a trout muscle sample recently collected in the Arctic yielded a chromatographic profde similar to that observed using negative ion chemical ionization mass spectrometry. A mixture of HCB and 2,4,dtrinitrotoluene(TNT) were shown to be distinguishable at 2.4-eV electron energy. The energetics of regioselective fragmentationof [4-l5N02FNT by dissociativeelectron capture can be determined on the fly. Complete negative ion gas chromatography/mass spectrometry (GC/MS) spectra were obtained for hexafluorobenzene and TNT by simultaneously ramping the electron energy from -2 to 15 eV and scanning the mass over a 200-Da range. Electron capture negative ion mass spectrometry (ECNIMS) is a sensitive and specific method for the analysis of electronegative compounds.l a 2 The method is particularly well suited for halogenated compounds, aromatic compounds, and nitro compounds, all of which are of concern as environmental pollutants. Standard ECNI-MS methods introduce a buffer gas into the ion source to moderate the electron energies necessary for resonance capture, but the introduction of the moderating gas produces artifacts that are difficult to control and that contribute adversely to the reproducibility of analytical results.3" A trochoidal electron monochromator attached to a quadrupole analyzer' recently was tested for the analysis of environmental chemicals.* This system produces slow electrons that are captured by electronegative compounds to give negative ions with high specificity and excellent sensitivity. It uses a magnetic field to confine the low-energy electrons (1) Dougherty, R. C. Anal. Chem. 1981, 53, 625636A.
(2) Hunt, D. F.; Crow, F. W. Anal. Chem. 1978,SO, 1781-1784. (3) Stbckl, D.; Budzikiewicz, H. Org. Mass. Spectrom. 1982, 27 (8), 376-381. (4) Stbckl, D.; Budzikiewicz, H.Org. Mass. Spectrom. 198517 (lo), 470-474. (5) Oehme, M.; Stbckl, D.; Knbppel. H. Anal. Chem. 1986,58, 554-558. (6)Stemmler, E. A.; Hites, R. A.; Arbogast, B.; Budde, W. L.; Deinzer, M. L.; Dougherty, R. C.; Eichelberger, J. W.; Foltz, R. L.;Grimm, C.; Grimsrude, E. P.;Sakashita, C.; Sears, L. J. Anal. Chem. 1988,60, 781-787. (7) Illenberger, E.; Scheunemann, H.-U.; BaumgBrtel, H. Chem. Phys. 1979.37, 12-31. (8) Laramb, J. A.; Kocher, C. A.; Deinzer, M. L. Anal. Chem. 1992.64.23162322.
0003-2700/94/036607 19$04.50/0 0 1994 American Chemical Society
and crossed electric and magnetic fields to disperse the electrons of different energies. The use of the electron monochromator has some distinct advantages over the standard practice of moderating high-energy electrons from the emitter with a buffer gas. Negative ions are generated when electrophilic molecules capture slow electrons. These ions result from capture of electrons in one or more resonance states, and thus unique energy spectra may be obtained for any given compound. The information is potentially useful for specific isomer identification and for confirmation of electron-capturing compounds. Other advantages are anticipated from this device, but most of them can be realized only with additional modifications in the instrument; the most important of immediate interest is the addition of a gas chromatographic inlet system. This has now been accomplished, and we report on the performance of this new system.
EXPERIMENTAL SECTION Gas Chromatography and CC/MS Interface. A HewlettPackard 57 10A gas chromatograph was connected to the trochoidal electron monochromator/mass spectrometer systems by an in-housedesigned heated transfer line. The transfer line consists of a 1.6-mm-diameter tube of 304 stainless steel which is surrounded by a 6.4-mm-0.d. tube togive it mechanical strength. Samples were introduced into the instrument by splitless injection through a DB-5 30 M X 0.25 mm i.d. capillary gas chromatography column with a 0.25-pm film thickness. Injector temperatures ranged from 150 to 250 OC, depending on the sample. The GC column was heated from 50 to 200 OC at 32 OC/min, and the transfer line was maintained at 230 OC. Helium flow rate was 55 mL/min. Electron Monochromator/Mass Spectrometer System. Electron energy measurements and mass analysis were made on an electron monochromator/mass spectrometer system which was described previously.s Mass spectra were acquired at fixed electron energy by ensemble averaging of approximately 15 mass scans across the GC peak at 325 amu/s-the fastest scan speed of the instrument. Energy spectra were also acquired across the GC peak by scanning the filament potential from +2 to -18 V over time intervals of 3-300 ms while recording either total ion currents or mass-resolved ion currents. Broad-band electron energies were produced by a Wavetek Model 190 function generator and used to acquire Anatyfical Chemistry, Voi. 66,No. 5, March 1, 1994 719
complete mass spectra of compounds eluting from the gas chromatograph. Electron energies were ramped from -2 to +15 eV at a rate that was approximately 10 times greater than the mass scan rate. Chemicals. Arctic trout muscle, which was purified to a nonpolar lipid-freefraction, was donated by the Environmental Protection Agency. A sample of pJ5NOz-labeled 2,4,6trinitrotoluene was provided by the U.S. Army ARDEC. Hexachlorobenzene was recrystallized three times and was better than 99% pure. Aroclor 1254 was obtained from Chemical Services, Inc. (West Chester, PA). Deconvolution. Spectra were deconvoluted into Gaussian peaks using the nonlinear curve fitting software PeakFit (Jandel Scientific,San Rafael, CA). The program was allowed free selection of the peak heights, peak positions, and peak widths. Peak widths chosen for deconvoluting the data in Figure 1 were restricted to one equal and common value which in turn was selected through the program software.
RESULTS The effluent from the capillary column was introduced directly into the ion chamber of the electron monochromator/ mass spectrometer where slow electrons from the monochromator mix with the analytes. Hexachlorobenzene (HCB) served as a standard to test the system. HCB captures electrons of 0.03-eV energy to produce an ion current that is composed of 97% molecular radical anion and 3% chloride ion. Sample amounts analyzed were in the range 5 pg to 10.8 ng. Excellent signal-to-noise ratios were observed even at lower concentrations (Figure la). Further reduction in sample amounts was not attempted. Narrow-range mass scanning about the molecular ion cluster of HCB showed mass resolution sufficient to separate the mass peaks in the isotopic cluster (Figure 1b). The mass range 275-295 amu was scanned at the instrument’s maximum rateof 325 amu/s. Hence, the totalduty cycle was well within the time constant of the gas chromatographic peak width of about 1.5 s. Better mass resolution is possible, but with the present source design this would be achieved at the expense of sensitivity. Nonlinear deconvolution of unresolved ion peaks could be useful when sensitivity limitations preclude full mass resolution. This is illustrated by deconvolution of the unresolved HCB molecular ion peak into the correct chlorine isotope pattern (Figure IC). Measurement accuracy of the chlorine isotope pattern has improved from 9% (Figure lb) to 6% (Figure IC) relative error. No systematic attempt was made to determine the reproducibility of the system. However, in early efforts to establish baseline information for performance of the instrument, HCB sample amounts injected into the gas chromatograph were recorded. Lower sample amounts (Figure 2) show improved response versus the amount of sample injected, while larger quantities show considerable scatter in the data. The response curve was expected to be flat over the range of injected sample amounts. The observed behavior suggeststhere are insufficient electrons available to reproducibly ionize the analyte molecules. The general utility of the system for GC/MS analyses was demonstrated by injecting varying amounts of Aroclor 1254 samples into the gas chromatograph and monitoring either the chloride ion or the total ion current. Injections of 45-190 720
Anatytical Chemistry, Voi. 66, No. 5, March 1. 1994
801
.. 410
351
uvv
415
. . -.-- ....--.-- . .. -* *.
420 425 GC time (sec)
.*
.*-a
-
*.
5
430
I
I
Figure 1. Ion count rate for the electron capture negative ion mass spectrometry of hexachlorobenrene(HCB) at 0.03-eV electron energy using a gas chromatography/eiectron monochromator/quadrupole mass spectrometer (GC/EM/MS) system with a 0.25 mm X 30 M DE5 column: (a) Ion chromatogram for the molecular ion peak of 5 pg of HCB. (b) Molecular anion cluster scan for 10.8 ng of HCB eluent from the capillary column. Relative error is 9%. (c) Anion current for 10.8 ng of HCB deconvolutedby Peakfit algorithm to give underlyingchiorlne isotope pattern. Relatlve error is 6 % .
ng of sample were used to produce ion chromatograms with 0.03-eV energy electrons. The sample amount injected to monitor the chloride ion was 45 ng (Figure 3a,b). An improved signal-to-noiseratio was observed when monitoring for chloride ion at 0.03-eV electron energy, as opposed to monitoring for total ion current, because background noise from other ionizable material in the source is reduced. These gas chromatograms compare very well with the patterns of negative ion mass spectra obtained on the same Aroclor mixture, and with the published spectrum of Aroclor 1254.9
22001 \
:.
fi
150'
5
100-
e
75-
0
*I
,I I :: .. :
4
m
w1
10
100 1000 let04 Amount Injected (pg)
let05
: r
"0
Figure 2. Response of the GC/EM/MS system as measured by the amount of molecular lon produced at 0.03-eV electron energy by hexachlorobenzene and normalized to amount inJected.
Application to environmental studies was demonstrated using a fish extract from an on-going research project for global deposition of environmental chemicals in the Arctic (Figure 3c). The sample was introduced through the gas chromatograph and monitored for chloride ion at 0.03 eV. The presence of many environmentally persistent chemicals is revealed in this nonpolar lipid-free fraction of the sample. These compounds had already been identified by standard chemical ionization negative ion mass spectrometry and include several isomers of tetrachloronaphthalene (TCN), pentachloronaphthalene (PCN), and polychlorobiphenyl(PCB) congeners and isomers with four to eight chlorine atoms. There are essentially three dimensions used in GC/MS studies for distinguishing closely related compounds. These are the retention times, ion mass peaks, and ion mass peak intensities. The capacity to generate discrete energy electron beams provides the opportunity to exploit a fourth dimension. All molecules that capture electrons to form either molecular or fragment ions have virtual orbitals that are unique, and most compounds have several states from which stable molecular radical anions or fragment anions are produced. The energies of these states are different for each molecule, and the electron energies for producing resonance ions from these states likewise are unique. Thus, isomers may be distinguished and compounds that are chromatographically unresolved may be identified, by their response to different electron energies. This principle is illustrated by the twocomponent mixture consisting of hexachlorobenzeneand 2,4,6trinitrotoluene (TNT) (Figure 4) which, although well separated by the high-resolution capillary column, nonetheless serve to demonstrate the principle of utilizing the added dimension of separation by electron energy. Scanning of the electron energy for hexachlorobenzene showed an intense peak at 0.03 eV. A similar electron energy scan for TNT gave a peak at 0.17 eV and another one with a maximum at 3.2 eV. When the analysis was carried out at 0.03 eV, both compounds showed chromatographic peaks (Figure 4a), but when the analysis was carried out at higher electron energies, the HCB peak intensity diminished, and at 2.4 eV the HCB peak was barely visible while the T N T peak dominated the chromatogram (Figure 4b). It should be noted that the sensitivity for (9) Erickson, M. D. Analytical Chemistry ofPCB's, Butterworth Boston, 1986; Chapter 6.
V
500
t
1000
1500
GC time (sec)
2000
2:
.
600
800
1000
GC time (sec)
1:
m
2
0
25
200
GC time (sec) Figure 3. Negathre Ion mass spectral analysis of polychlorlnated compounds using the gas chromatography/electronmonochromator/ mass spectrometer instrument at 0.03-eV electron energy. A 0.25 mm X 30 M DB-5 capillary column was used for chromatographlc separation: (a) Chloride lon chromatogram of 45 ng of Arocbr 1254 mlxture. Normalized to 100%. (b) Ion count rate for expanded chromatogram from 600 to 1200 s showing welkesolved peaks. (c) Trout muscle sample from the Arctic. Normallred to 100%.
TNT detection at 2.4 eV is a factor of 13.5 lower than it is at 0.03 eV (compare panels a and b of Figure 4), which exactly agrees with measured differences in peak intensities when the electron energies are scanned. This simple demonstration shows how compound identification or confirmation could be facilitated with another physically observable variable. As a rule, the most intense molecular ions of nearly all electrophilic compounds occur near 0 eV. This is a consequence of the Wigner threshold law. Fragment ion formation resulting from dissociative electron capture usually has higher energy requirements. These may be anywhere from 0 to 6 eV Anelytical Chemlstty, Vol. 66, No. 5, March 1, lQQ4
721
JUUU
1.5&04
% m
1.25e+04-
8
1
(a)
le04-
4
m
7500 -
2
..
0
. ..
.e
50001
.. .
1
._ e
.
2500
0 440
900-
%
800-
-l! 700-
a!
460
(b)
500
520
GC tune (sec)
I
..
..
N
600500d 2 4.004
m v)
100
0
440
*
*
.. ... *.. rL.
460
480 GC time (sec)
500
I
m
Flgure 4. Total ion chromatogram of a 2,4,6-trlnttrotoluene (21.2 ng) and hexachbrobenzene(5.4 ng) mixtureusingthe gas chromatography/ electron monochromator/massspectrometrysystem: (a) Total number of anlons produced at (a) 0.03-eV electron energy and (b) at 2.4-eV electron energy.
or higher. Accordingly, to obtain complete mass spectra in the negative ion mode from gas chromatographic effluents, a constant population of electrons over this energy range is required. There are several ways this can be accomplished. One is to enlarge the lens apertures. This would increase the electron energy range as well as the electron flux, thereby improving the sensitivity. But it would reduce the electron energy resolution. A more practical approach is to ramp the electron energies at high frequency over the range needed for the analysis. Hexafluorobenzene has a single resonance state for the molecular ion and a peak maximum is observed at 0.03-eV electron energy. A resonance state for production of C6F5ion requires 4.5-eV electron energy. When the electron energy was scanned from-3 to +9 eV in 3 ms, two peaks were observed (Figure 5a).1° The slew rates over this energy range were varied from 20.8 to 2080 eV/s, and the signal for C6F6- ion was measured (Figure 5b). The signal intensity diminished at higher slew rates, but even at 500 eV/s the response was within 75% of maximum. These results confirm that the rate at which the electron energies are scanned will not be a limitation for analyzing gas chromatographic effluents. Thus, 50 scans are possible over a 2-s-wide gas chromatographic effluent peak when a slew rate of 250 eV/s is used to scan a 10-eV energy range. At this rate, the signal intensity would be better than 90% of maximum (Figure 5b). (10) There also is a weak state at 8.3 eV for formation of the C6F5which did not show up under rapid energy scanning.
722
An8!ytical Chemlstty, Vol. 66, No. 5, March 1, 1994
500
1000
1500
Slew Rate (volt/sec)
2000
2
Flgurr 5. (a) Electron energy scan from -3 to +9 eV for hexafluorobenzene completed in 3 ms. (b) RelativeCeFd molecular anion yield electrons) over the electron energy at dlfferent slew rates (for 0.0-V range -3 to +9 eV.
In a previous study with chlorine-37regiospecificallylabeled [ l-37C1],3-dichlorodioxin, it was shown that the two chlorines dissociate from the parent with distinctly different electron energies." This type of regioselectivity was further demonstrated by scanning ion peaks m / z 46 and 47 with respect to electron energy forp-l5NOz-1abeled TNT. When the sample was introduced through the solid inlet probe, the analyses showed that ortho and para nitro groups dissociate from the parent molecule with electrons of different energies (Figure 6). The major difference between the spectra is a strong resonance peak at 0.17 eV, for dissociationof thep-nitro group and a much weaker one at 0.13 eV for dissociation of o-nitro groups. Ion intensity electron energy profiles for dissociation of both the p- and o-nitro groups showed a second broad electron energy peak with maxima around 5 eV. Deconvolution by a nonlinear algorithm, PeakFit, suggested there are several negative ion resonance states in each. Computer-controlled electron energy scanning of a gas chromatographic effluent of the nitrogen-15-labeled compound showed similar electron energy profiles. An electron energy range of 20 eV typically was scanned in 0.3 s. The differences in these spectra are probably related to the differences in the exit lens apertures, which in the case of the probe samples was 0.5 mm and for the gas chromatography experiments was 1 mm. The precision in electron energy beams is fO.l eV in the former case and f0.5 eV in the latter. The larger lens aperture was necessary for the GC experiment to compensate for the reduced amount of sample introduced into the instrument. (1 1) Laramk, J. A,; Deinzer, M. L. The 40rh Conference on Mass Spectrometry
and Allied Topics May 31-June 5, 1992; p 995.
-5
0
5 IO 15 Electron Enerpy (eV)
20
Figure 8. Electronenergies for the dlssociathre electron capture reactlon paths of 2,[4-'6N0~],&trinitrotoluene. Spectra are normallzed to 100%: (a) Sample introduced via the solM Inlet probe. Lens aperture was 0.5 mm. (b) Sample Introduced by gas chromatography. Lens aperture was 1.0 mm.
400
350 al 300 0
9
a
250
5 200
2 g
150
w
100 50 0
80
120
I60
2
m/z
Figure 7. Gas chromatography/mass spectrometry with ramping of electron energy from -2 to 15 eV for (a) hexafluorobenzeneand (b) 2,4,&trlnltrotoluene. Several hundredmicrograms of sample was used In each case.
+
Rapid ramping of electron energies is essential for obtaining complete mass spectra of compounds with broadly centered resonance states. This was demonstrated by GC/MS experiments carried out with hexafluorobenzene and TNT (Figure 7). Hexafluorobenzene (Figure 7a) is particularly well suited for this purpose because each ion is produced from one or more unique resonance states. Thus, the molecular radical anion shows a peak maximum with 0.03-eV energy
electrons, the C6Fs- ion with 4.5- and 8.3-eV electrons, and the F ion with 4.5- and 10-eV electrons. There is no overlap of resonances for production of the molecular ion and fragment ions; therefore, high-frequency ramping of the electron energy is necessary to obtain the complete mass spectrum. Rapid energy scanning of TNT likewise resulted in a mass spectrum with a largenumber of ion peaks (Figure 7b), but themolecular radical anion was not recorded due to the nonsynchronized scanning method. Although the sensitivity in these experiments was not impressive (several hundred micrograms of sample was introduced through the GC), it is clear that completenegative ion spectra of electron-capturing compounds can be obtained with this system.
DISCUSSION The experiments described here show that the electron monochromator/mass spectrometer equipped with a gas chromatograph can be used for negative ion GC/MS analyses of environmental compounds. This modification greatly enhances the system for such analyses, and with further improvements, an analytical instrument should become available that eliminates some of the problems encountered with the use of buffer gases for production of slow electrons. Moreover, the system has the potential for providing another dimension for the identification and confirmation of environmental chemicals. Each compound has unique resonance states from which molecular ion or product ion intensities can be maximized by selecting electrons with the right energies. These states and the electron energies required to produce the ions can be predicted by semiempirical or ab initio calculations which could be useful for the identification of compounds when standards are not available. Ultimately, two-dimensional energy/mass scanning could provide a fingerprint for any electron-capturing compound, isomer, or congener. In the present prototype, sensitivity is limited by several factors. First, the electron beam emerges from an aperture of 0.5" diameter. On examination of the first lens in the lens stack, a perceptible electron burn is observed which is 6 mm in diameter and this constitutes an estimated 90% of the AMlytlcalChemistry, Vol. 66, NO. 5, Msrch 1, 1884
723
total electron current. Thus, the fraction of the beam used for ionization is only 1 part in 144. By increasing the size of the aperture a significantly greater fraction of the electron beam becomes available resulting in greatly improved sensitivity, but at the expense of electron energy resolution. This can be compensated for by appropriate adjustment of the electron drift distance, magnetic field strength, which also controls brightness of the electron beam, and electric field strength.12 Second, emitters with low work functions can be used to produce more intense electron beams at lower operating temperature^.'^ The work functions for lanthanum hexaboride (2.6 eV) and cerium hexaboride (2.4 eV), for example, are considerably lower then that of tungsten (4.5 eV). And indirectly heated cathodes can readily produce 10-mAelectron currents.14 The space charging effect also limits the usable electron current, and this can be corrected by using Pierce elements or Wehnelt CUSPS.'~ Third, the purpose of the standard chemical ionization reagent gas in the source for production of negative ions is twofold. First, the reagent gas acts to generate slow electrons. Second, once formed, the ions, particularly the molecular ions, are stabilized against autodetachment of the electron through ~~~
ACKNOWLEDGMENT This work was supported by the National Institutes of Health (Grants ES 00040 and ES 00210). This is Technical Paper 10293 from the Oregon Agricultural Experiment Station. Helpful comments from Donald Griffin are gratefully acknowledged. We also thank Steve Sonnen for assistance in the analysis.
~~
(12) Stamatovic, A; Schulz, G. J . Reu. Sci. Imrrum. 1970, 41, 4 2 3 4 2 7 . (13) Davis, P. R.; Schwind, G. A. Appl. SurJ Sci. 1986, 25, 355-363. (14) Boumsellek, S.; Chutjian, A. A m l . Chem. 1992, 64, 2096-2100. (15) Hawkes, D. W.; Kasper, E. Principles of Elecfron Oprics; Academic Press: London, 1989; Vol. 2.
724
collisionswith the neutral buffer gas molecules. This improves the detection sensitivity. The introduction of a buffer gas into the ion source whose only purpose is to stabilize the ions should improve the performance of the electron monochromator/mass spectrometer. The buffer gas selected should not ionize under the experimental conditions; therefore, it would preferably have a high ionization potential, low electron affinity, and low chemical reactivity. Additionally sufficient vibrational modes would be necessary to efficiently quench excess energy in the anion. The gas pressures in the ion source, however, would be adjusted to stabilize the molecular ions, not to produce electrons with the right energies. This should involve gas pressures that are lower than those typically used in chemical ionization experiments. With such a system the electron energies would be independent of the gas pressures and stabilization energies for the ions could be determined.
Ana&ficalChem/stty, Vol. 66, No. 5, March 1, 1994
Recehred for review September 20, 1993. Accepted December 15, 1993." ~~
e
Abstract published in Aduance ACS Abstracts, February 1 , 1994.
